The Association of Physical Activity Fragmentation with Physical Function in Older Adults: Analysis from the SITLESS Study

Round 1

Reviewer 1 Report

This paper has been an interesting read. It must have been a hard work to analyze actigraph data from more than 1000 people. The article is well written and presented. However, there are some small mistakes than must be corrected. Furthermore, the paper would gain some quality if discussion was improved.

Line 77. ASTP has been previously described in the abstract.

Line 35 and 104. Author could explain the meaning of the name SITLESS.

Line 112. SPPB has been previously described in the abstract. 

Line 123. Even the study procedures are described somewhere else, authors must define it in this paper as well

Line 133. There is a mistake in the references 25-27. Authors used two different kind of brackets [ ). Correct this.

The main limitation of this paper to be published in this journal is that the discussion is poor compared to the overall merit of the paper. It must be improved in general. Mostly, I would be important to deepen on the discussion about SPPB and 2MWT as they are direct measurements done on the sample.

 

 

Author Response

This paper has been an interesting read. It must have been a hard work to analyze actigraph data from more than 1000 people. The article is well written and presented. However, there are some small mistakes than must be corrected. Furthermore, the paper would gain some quality if discussion was improved.

Response: Thank you for this comment, hopefully we have sufficiently addressed your queries in the following points.

 

Line 77. ASTP has been previously described in the abstract.

Response: It is usually good practice to also fully define the first mention of any acronyms within the manuscript text as some readers may not initially look at the abstract to see the original use of the acronym.

 

Line 35 and 104. Author could explain the meaning of the name SITLESS.

Response: SITLESS is simply the name of the project, it does not have a fuller meaning.

 

Line 112. SPPB has been previously described in the abstract.

Response: It is usually good practice to also fully define the first mention of any acronyms within the manuscript text as some readers may not initially look at the abstract to see the original use of the acronym.

 

Line 123. Even the study procedures are described somewhere else, authors must define it in this paper as well

Response: Sections 2.2, 2.3 and 2.4 fully highlight the relevant study procedures for this particular paper; Reference 22 covers other aspects of the SITLESS project which do not necessarily add anything to this particular paper.

 

Line 133. There is a mistake in the references 25-27. Authors used two different kind of brackets [ ). Correct this.

Response: Well-spotted, this has now been updated to [25-27].

 

The main limitation of this paper to be published in this journal is that the discussion is poor compared to the overall merit of the paper. It must be improved in general. Mostly, I would be important to deepen on the discussion about SPPB and 2MWT as they are direct measurements done on the sample.

Response: This comment is a little vague but we have tried to address it by adding some extra sentences to strengthen the Discussion section. If you still feel this has not been clarified, then please feel free to be more specific on what you would like addressed and we will try to oblige.

Reviewer 2 Report

It is considered very meaningful as a study using a new protocol for physical activity of the elderly. In particular, since it is an accessible protocol, I expect continued research results.

Just a few comments.

 

1. In introduction,

“Never-the-less, it has been suggested that a more fragmented physical activity pattern with many short physical activity bouts may suggest higher levels of frailty and fatigability [12,13].”

This sentence is ambiguous. From the above content that the existing protocol is changing to a new protocol, suddenly the opposite appears, but the logical flow is rather awkward.

 

2. In introduction,

However, these minimum bout lengths have currently been untested. This aspect is important to consider as the time spent in different bouts is likely to reflect a physiological response to a treatment or an intervention.

What does physiological response mean? It is not clear because it can be interpreted in so many different ways.

 

3. As a study using data from The SITLESS study, please add the period of The SITLESS study or the period during which the data used in this study was collected.

Depending on the duration of the study, if there was an effect of COVID-19, a part about how it was handled should be added.

 

4. Also, please add the process of extracting the data of this study from The SITLESS study.

Was the number of subjects unchanged from 1360 from the beginning of the program? Were there any subjects dropped or excluded during the study? What is expressed in Table 4 has been confirmed. However, it seems that information related to subject screening is needed in the subject part.

 

5. Tables S1-S10 are too complex in Supplementary Material.

Please organize them for better readability.

Also consider whether the t-value must be written.

 

6. In discussion,

Lower ASTP using ≥120-second and ≥300-second physical activity bouts resulted in longer 2MWT distances although no significant associations were found for ASTP using  ≥60-second physical activity bouts. Our study has continued to add weight to the idea that more fragmented physical activity patterns may be an indirect indicator of reduced functional capacity in older adults.

Although it is not statistically significant in this sentence, please write the significance of the increased 2MWT distance with other papers. Also, the two sentences are not logically connected.

 

7. What is the rationale for higher thresholds (e.g. ≥300-seconds)?

 

Author Response

It is considered very meaningful as a study using a new protocol for physical activity of the elderly. In particular, since it is an accessible protocol, I expect continued research results.

Response: Thank you for this comment.

 

1. In introduction,

“Never-the-less, it has been suggested that a more fragmented physical activity pattern with many short physical activity bouts may suggest higher levels of frailty and fatigability [12,13].” This sentence is ambiguous. From the above content that the existing protocol is changing to a new protocol, suddenly the opposite appears, but the logical flow is rather awkward.

Response: We have now updated this sentence to provide a better flow and also a better explanation of the key terms (Lines 69-72).

 

2. In introduction, However, these minimum bout lengths have currently been untested. This aspect is important to consider as the time spent in different bouts is likely to reflect a physiological response to a treatment or an intervention. What does physiological response mean? It is not clear because it can be interpreted in so many different ways.

Response: This sentence has now been deleted as it was ambiguous.

 

3. As a study using data from The SITLESS study, please add the period of The SITLESS study or the period during which the data used in this study was collected. Depending on the duration of the study, if there was an effect of COVID-19, a part about how it was handled should be added.

Response: Clarification has now been added that this current paper uses a cross-sectional design using baseline data from the SITLESS study which was collected from July 2016 to December 2017. This highlights it was not impacted by the COVID-19 pandemic.

 

4. Also, please add the process of extracting the data of this study from The SITLESS study. Was the number of subjects unchanged from 1360 from the beginning of the program? Were there any subjects dropped or excluded during the study? What is expressed in Table 4 has been confirmed. However, it seems that information related to subject screening is needed in the subject part.

Response: The current study used a cross-sectional design using baseline data from the SITLESS study (clarified at the beginning of the Methods section). The reasons for missing data (e.g. not providing valid accelerometry data) have been highlighted in the Results section (Lines 210-213).

 

5. Tables S1-S10 are too complex in Supplementary Material. Please organize them for better readability. Also consider whether the t-value must be written.

Response: The supplementary materials were developed alongside co-author Dr Ilona McMullan who is a statistician. Best practice guidelines were followed for reporting the results of linear regressions. If the editor wants us remove these then we can do so.

 

6. In discussion,

Lower ASTP using ≥120-second and ≥300-second physical activity bouts resulted in longer 2MWT distances although no significant associations were found for ASTP using  ≥60-second physical activity bouts. Our study has continued to add weight to the idea that more fragmented physical activity patterns may be an indirect indicator of reduced functional capacity in older adults. Although it is not statistically significant in this sentence, please write the significance of the increased 2MWT distance with other papers. Also, the two sentences are not logically connected.

Response: The particular paper referenced did not use the 2MWT but the 6MWT instead (a similar walking test). This has now been highlighted and the level of significance as well. The two sentences are also now more coherent.

 

7. What is the rationale for higher thresholds (e.g. ≥300-seconds)?

Response: The justification for exploring higher thresholds has been provided in the Introduction (see lines 92-96) and Discussion (see lines 285-288).

Reviewer 3 Report

The authors investigated the associations of physical activity bouts of elderlies with physical function (2-minute walk and short physical performance battery) and explored which minimum length of bouts used for ASTP was significantly associated. The authors showed and suggested that ≥10-second bouts predicted physical function comparing with longer minimum bouts. A large sample size and multi-center data strengthen the results. The viewpoints are interesting, but there are concerns about the study design.

 

1. Epochs of accelerometers for the elderly are usually set to 60 seconds. References 15 and 16 the authors cited used the 60-second epochs. The review (Int J Behav Nutr Phys Act. 2018 Nov 26;15(1):119. doi: 10.1186/s12966-018-0749-2) suggested that using 60-second epochs was most valid for elderly. However, the authors cited Reference 19, which described child physical activity. For children, 10-second- or less length epochs are commonly used. If 60-second epochs are used, ≥10-second bouts are not available. The authors should delete Ref. 19 and explain why ≥10-second bouts are adopted for the elderly citing appropriate references.

 

2. The active-to-sedentary transitioning probability (ASTP) was defined in Ref. 15. But, why this is a probability is not explained in Ref. 15. The ASTP is the reciprocal of the average physical activity bout duration. Daily physical activity bout duration ranges 0 to19 hour (1140 minutes), so the ASTP ranges near 0 to +infinite. The authors should add more explanation about the ASTP.

 

3. The ASTP is the reciprocal of the average physical activity bout duration. The authors expressed the ASTP as percentages in Table 3. The authors should describe what the total (100%) indicates in the Materials and Methods.

 

4. The effect size of Total PA (using 10-second epochs) is the largest in Table 4 (standardized beta coefficients). It is unknown whether the ASTP is valuable. The authors did not mention total PA in the Results and Discussion.

 

5. According to Ref. 15, when using each minimum bouts, corresponding epochs should be used. For example, when using ≥60-sedcond bouts, 60-second epochs should be used, and when using ≥120-second bouts, 120-second epochs should be used, and so on. Could you show the results from this way in sensitivity analyses?

 

6. Lines 148–1150 “In order to provide more comparable ASTP scores across the four different minimum bout lengths, the ≥10-second, ≥120-second and ≥300-second bout lengths were normalised to the ≥60-second data.” This standardization maneuvers (this may not be normal-distribution transformation?) do not influence the results in regression analyses, that is, standardized beta coefficients, and p values. However, the authors did not mention comparison of the ASTP values among different minimum bout lengths in the Results and Discussion. First, the authors should explain comparability in the Materials and Methods, and then describe the comparison in the Results, and discussed it in the Discussion section.

 

Minor points.

7. Ref. 16 does not use the ASTP. Line 78.

 

8. The authors used “hierarchical linear regression analysis.” Line 187. What variables are used for clusters? All variables used in the models are shown as fixed effect variables in Supplementary Table. Explain it in the Methods.

 

9. Commas in 187, and 245 should be deleted.

Author Response

The authors investigated the associations of physical activity bouts of elderlies with physical function (2-minute walk and short physical performance battery) and explored which minimum length of bouts used for ASTP was significantly associated. The authors showed and suggested that ≥10-second bouts predicted physical function comparing with longer minimum bouts. A large sample size and multi-center data strengthen the results. The viewpoints are interesting, but there are concerns about the study design.

Response: Thank you for this comment, hopefully we have sufficiently addressed your queries in the following points.

 

Epochs of accelerometers for the elderly are usually set to 60 seconds. References 15 and 16 the authors cited used the 60-second epochs. The review (Int J Behav Nutr Phys Act. 2018 Nov 26;15(1):119. doi: 10.1186/s12966-018-0749-2) suggested that using 60-second epochs was most valid for elderly. However, the authors cited Reference 19, which described child physical activity. For children, 10-second- or less length epochs are commonly used. If 60-second epochs are used, ≥10-second bouts are not available. The authors should delete Ref. 19 and explain why ≥10-second bouts are adopted for the elderly citing appropriate references.

Response: We agree including this particular reference was not appropriate and we have now replaced it with a suitable older adult example. We have also added clarification in the Methods section that the ActiLife software automatically linearly scaled down the cut point threshold of 100 cpm accordingly to adjust for the 10-second epoch length.  To clarify, despite 60-second epochs being commonly used in research, we feel this is a historical artifact based off the previous limitations of early accelerometers which were unable to assess accelerometry using lower epoch lengths. There is still ambiguity within the research literature about what is the most appropriate epoch length to measure physical activity in older adults.

 

The active-to-sedentary transitioning probability (ASTP) was defined in Ref. 15. But, why this is a probability is not explained in Ref. 15. The ASTP is the reciprocal of the average physical activity bout duration. Daily physical activity bout duration ranges 0 to19 hour (1140 minutes), so the ASTP ranges near 0 to +infinite. The authors should add more explanation about the ASTP.

Response: We have added further explanation that the ASTP is defined as the probability of transitioning from an active state to sedentary state and also added further explanation in answering your query below.

 

The ASTP is the reciprocal of the average physical activity bout duration. The authors expressed the ASTP as percentages in Table 3. The authors should describe what the total (100%) indicates in the Materials and Methods.

Response: We have now clarified that an ASTP value of 100% would suggest that, on average, an individual only accrues their bouts of physical activity at the minimum bout length threshold.

 

The effect size of Total PA (using 10-second epochs) is the largest in Table 4 (standardized beta coefficients). It is unknown whether the ASTP is valuable. The authors did not mention total PA in the Results and Discussion.

Response: We only briefly refer to the association between total physical activity and physical function in older adults in the first paragraph of the Discussion section as the main focus of this paper was on the potential usefulness of assessing physical activity fragmentation as an indirect method of determining physical function in older adults. There has also been plenty of research highlighting how increased time spent in total physical activity leads to improved physical function so we did not want to labour this point.

 

According to Ref. 15, when using each minimum bouts, corresponding epochs should be used. For example, when using ≥60-sedcond bouts, 60-second epochs should be used, and when using ≥120-second bouts, 120-second epochs should be used, and so on. Could you show the results from this way in sensitivity analyses?

Response: This is certainly an interesting point. However, we feel that with the amount of analyses already contained within this current paper and the fact another reviewer commented about the complexity of the presented analyses that adding further to this by pairing epoch length with bout length may make the key messages contained within this current paper less clear. However, it is an important point to mention so we have added this to the Conclusion section as a possible area for future research.

 

Lines 148–150 “In order to provide more comparable ASTP scores across the four different minimum bout lengths, the ≥10-second, ≥120-second and ≥300-second bout lengths were normalised to the ≥60-second data.” This standardization maneuvers (this may not be normal-distribution transformation?) do not influence the results in regression analyses, that is, standardized beta coefficients, and p values. However, the authors did not mention comparison of the ASTP values among different minimum bout lengths in the Results and Discussion. First, the authors should explain comparability in the Materials and Methods, and then describe the comparison in the Results, and discussed it in the Discussion section.

Response: As you correctly highlight, the normalising process was simply to aid comparison between the different minimum bout lengths utilised and do not alter any of the observed associations found. However, there have already been comments from one of the other reviewers about the complexity and amount of analyses being completed so we feel adding anything extra would distract from the key messages this particular paper is trying to make.

 

Minor points.

Ref. 16 does not use the ASTP. Line 78.

Response: This reference has now been removed from Line 78.

 

The authors used “hierarchical linear regression analysis.” Line 187. What variables are used for clusters? All variables used in the models are shown as fixed effect variables in Supplementary Table. Explain it in the Methods.

Response: We can clarify that we adjusted the analysis by country to account for clustering effect. This has now been added to the Methods section.

 

Commas in 187, and 245 should be deleted.

Response: These commas have now been removed.

Round 2

Reviewer 1 Report

Authors addressed my concerns and it is now suitable for publication

Author Response

Again, thank you for your comments which helped to improve our manuscript.

Reviewer 3 Report

Please see an attached file. The definition of an ASTP is confused.

Comments for author File: Comments.pdf

Author Response

A reply to your queries has been attached.

Author Response File: Author Response.docx